Catherine Laporte

Swallowing is a critical marker of neurological and emotional health. The ability to monitor it continuously and non-invasively, especially through smart ear-worn devices, holds significant promise for clinical applications. Despite this potential, no public audio datasets currently support reliable swallowing sound detection. Existing datasets focus primarily on speech and breathing, offering limited coverage and lacking detailed annotations for swallowing events. To address this gap, we introduce an in-ear audio dataset specifically designed to capture a wide range of verbal and non-verbal sounds. It includes comprehensive labeling focused on swallowing. The dataset was collected from 34 healthy adults (14 females and 20 males) between the ages of 20 and 29. Each participant performed a series of predefined tasks involving both non-verbal and verbal events. Non-verbal tasks included swallowing, clicking, forceful blinking, touching the scalp, and physical movements such as squatting or walking in place. Verbal tasks consisted of speaking (e.g., describing an image). Recordings were conducted in both quiet and noisy environments to better reflect real-world conditions. Data were captured using a combination of in-/outer-ear microphones, a chest belt to record electrocardiogram (ECG), respiration and acceleration signals, and an ultrasound probe to track tongue movement, which served as a reference for swallowing annotation. All signals were precisely synchronized. To ensure high data quality, the recordings were reviewed using both algorithmic analysis and manual inspection. Swallowing events were identified based on ultrasound signals and validated by an expert to guarantee accurate labeling. As a proof of concept that in-ear audio supports swallow classification, we fine-tune a fully connected neural network on YAMNet embeddings plus zero-crossing rate (ZCR) features. Across the completed folds, the model reaches an F1 score of 0.875 ± 0.013.

Freehand 3D ultrasound imaging is emerging as a promising modality for regular spine exams due to its non-invasiveness and affordability. The laminae landmarks play a critical role in depicting the 3D shape of the spine. However, the extraction of the 3D lamina curves from transverse ultrasound sequences presents a challenging task, primarily attributed to the presence of diverse contrast variations, imaging artifacts, the complex surface of vertebral bones, and the difficulties associated with probe manipulation. This paper proposes Sequential Localization Recurrent Convolutional Networks (SL-RCN), a novel deep learning model that takes the contextual relationships into account and embeds the transformation matrix feature as a 3D knowledge base to enhance accurate ultrasound sequence analysis. The assessment involved the analysis of 3D ultrasound sequences obtained from 10 healthy adult human participants, covering both the lumbar and thoracic regions. The performance of SL-RCN is evaluated through 7-fold cross-validation, employing the leave-one-participant-out strategy. The validity of the AI model training is assessed on test data from 3 participants. Normalized Discrete Fréchet Distance (NDFD) is employed as the main metric to evaluate the disparity of the extracted 3D lamina curves. In contrast to our previous 2D image analysis method, SL-RCN generates reduced left/right mean distance errors from 1.62/1.63mm to 1.41/1.40mm, and NDFDs from 0.5910/0.6389 to 0.4276/0.4567. The increase in the mean NDFD value from 7-fold cross-validation to the test-data experiment is less than 0.05. The experiments demonstrate the SL-RCN’s capability in extracting accurate paired smooth lamina landmark curves.

Goal: Uncertainty estimations through approximate Bayesian inference provide interesting insights to deep neural networks' behavior. In unsupervised learning tasks, where expert labels are unavailable, it becomes ever more important to critique the model through uncertainties. Methods: This paper presents a proof-of-concept for generalizing the aleatoric and epistemic uncertainties in unsupervised MR-CT synthesis of scoliotic spines. A novel adaptation of the cycle-consistency constraint in CycleGAN is proposed such that the model predicts the aleatoric uncertainty maps in addition to the standard volume-to-volume translation between Magnetic Resonance (MR) and Computed Tomography (CT) data. Results: Ablation experiments were performed to understand uncertainty estimation as an implicit regularizer and a measure of the model's confidence. Conclusion: The aleatoric uncertainty helps in distinguishing between the bone and soft-tissue regions in CT and MR data during translation, while the epistemic uncertainty provides interpretable information to the user for downstream tasks.

Transcranial Doppler ultrasound (TCD) is a method that uses a hand held low frequency (2-2.5 MHz), pulsed Doppler phased array probe to measure blood velocity within the arteries located inside the brain. The problem with TCD lies in the low ultrasonic energy penetrating inside the brain through the skull which leads to low signal to noise ratio. This is due to several effects including phase aberration, variations in the speed of sound in the skull, scattering, the acoustic impedance mismatch and absorption of the three layer medium constituted by soft tissues, the skull and the brain. The goal of this paper is to study the effect of transmission losses due to the acoustic impedance mismatch on the transmitted energies as a function of frequency. To do so, wave propagation was modelled from the ultrasonic transducer into the brain. This model calculates transmission coefficients inside the brain, leading to a frequency-dependent transmission coefficient for a given skin and bone thickness. This approach was validated experimentally by comparing the analytical results with measurements obtained from a bone phantom plate mimicking the skull. The average position error of the occurrence of the maximum amplitude between the experiment and analytical result was equivalent to a 0.06 mm error on the skin thickness given a fixed bone thickness. Similarity between the experimental and analytical result was also demonstrated by calculating correlation coefficients. The average correlation between the experimental and analytical result came out to be 0.50 for a high frequency probe and 0.78 for a lower frequency probe. Further analysis of the simulation showed that an optimized excitation frequency can be chosen based on skin and bone thicknesses, thereby offering an opportunity to improve the image quality of TCD.

The use of mechanically representative phantoms is important for experimental validation in ultrasound (US) imaging, elastography, and image registration. This article proposes a model to predict the elastic modulus of a soft tissue-mimicking phantom based on two very easily controllable parameters: gelatin concentration and refrigeration duration. The model has been validated on small- and large-scale phantoms; it provides a good prediction of the elastic modulus in both cases (error < 16.2%). The tissue-mimicking phantom is made following a low-cost and simple fabrication procedure using commercial household gelatin with psyllium hydrophilic mucilloid fiber to obtain echogenicity. A large range of elastic properties was obtained (15-100kPa) by adjusting the gelatin concentration between 5% and 20% (g/mL) and the refrigeration time of the sample between 2 and 168 h, allowing to mimic normal and pathological human soft tissues. The phantom's acoustic properties (velocity, attenuation, and acoustic impedance) are also assessed using the American Institute of Ultrasound in Medicine (AIUM) standard.